How TranscribeVid works

TranscribeVid turns the speech in a video or audio file into text with timestamps, in three simple steps.

3 simple steps

Add your file
Drop a video or audio file on the home page, or record from your microphone.
Pick quality
Choose Fastest, Standard, High accuracy or Best accuracy: Fastest is quickest, Best accuracy makes the fewest mistakes.
Read, edit, download
Watch the text appear live, fix any mistakes, then download it as SRT, VTT, TXT, Markdown, JSON or PDF.

Accurate text
Usually very accurate on clear speech. Lines it is unsure about are marked, so you can check them first.
Timestamps
Every line has a time. Play the media from any line to hear what was said.
More than 50 languages
Choose the spoken language or leave it on auto-detect. You can also translate speech into English.
Edit it right here
Change or delete a line, or search and replace. Your edits are in every download.
Six ways to download
SRT and VTT for subtitles, TXT and Markdown for reading, JSON for programs, or a PDF to print.
Private
Your file never leaves your device. It is transcribed right in your browser, with no account, and your transcripts stay in this browser.

Choose the spoken language instead of auto-detect.
Add a vocabulary hint: names, brands and jargon, spelled the way you want them.
Pick High accuracy or Best accuracy for difficult audio.
Re-transcribe with another quality or language; your current transcript is kept.
Start from good audio: little background noise, a close microphone, one person at a time. Keep "Enhance audio" on.
Check the marked lines and fix them in the transcript editor.

File limits. None set by us. Your file is processed on your device, so very large or long files take longer and need enough free memory; the page warns you on a slower device.
Speed. It depends on your device, the quality you pick and the length of the file. The first use of a quality level downloads its model once: your browser fetches the model files from Hugging Face and the engine's program files from jsDelivr, then keeps them cached. Keep the tab open while it works.
Nothing is saved for later. Transcripts and the file stay in this browser only and are deleted automatically after 30 minutes of inactivity, or when you press Delete my data now. Export what you need before you leave. Details are in the Privacy policy.
Ads. Ads appear only on information pages and beside finished transcripts, and are served by Google, which uses cookies. No ad is shown while a file is being transcribed.
Mistakes happen. No speech recognition is perfect. Always verify a transcript before relying on it for anything important.